The first section of these results draws from the results files generated with a 100kb window, stepped at 50kb intervals.

Firstly, the VCFtools Fst calculator emits both ‘raw’ and ‘adjusted’ Fst values, so let’s take a look at how closely these two measures predict one another…

The plot below shows the relationship of the two Fst estimates across all scaffolds and all 15 comparisons (the red line is 1:1). The R2 for the correlation here is 0.93, suggesting that it probably doesn’t matter a great deal which statistic we use – though the adjusted Fst is generally somewhat higher.

However, there does seem to be a difference when it comes to the consistency of Fst bewteen scaffolds; here illustrated with just (the SSR-bearing) scaffolds 4, 8, 22, 60 & 77.

## Using SCAF as id variables

Note the increase in crossing over and the greater variablility in some of the comparisons, particularly those involving BAM and BAVA.

Taking a slightly deeper look, lets see if we can get some idea of the variation within scaffolds – I’ll start with the SSR-bearing scaffolds for want of a more compelling place to start.

So in this plot we can see a few things; firstly, that the ordering of the comparisons doesn’t change an awful lot between or along these scaffolds. Secondly, there is noticeably more variability along scaffolds 60 & 77 than along the (longer) 4, 22 & 8.

To put that numerically the variance in Fst along these ranges from 1.310^{-4} to 0.042. The COB_MOV comparison is the most variable over all (mean var. of 0.0263881), and APA_BAM the least variable (mean var of 0.001071).


Section 2

For a quick comparison, I’m going to throw together the same plots using the results files generated with a 200kb window, stepped at 100kb intervals.

The first thing to note is that the R2 for the correlation between the two Fst measures is still 0.93.

## Using SCAF as id variables

Secondly, we can see that this plot of mean adjusted Fst is very similar to the one above…

…and thirdly, the pattern of Fst along the SSR-bearing scaffolds also looks pretty much the same. So, do we actually see differences between the results from differing window sizes anywhere?